Systematic Epistasis Analysis of the Contributions of ...performed comprehensive epistasis analysis...

Copyright � 2010 by the Genetics Society of AmericaDOI: 10.1534/genetics.110.115808

Systematic Epistasis Analysis of the Contributions of Protein KinaseA- and Mitogen-Activated Protein Kinase-Dependent Signaling

to Nutrient Limitation-Evoked Responses in the YeastSaccharomyces cerevisiae

Raymond E. Chen1 and Jeremy Thorner2

Division of Biochemistry and Molecular Biology, Department of Molecular and Cell Biology,University of California, Berkeley, California 94720-3202

Manuscript received February 20, 2010Accepted for publication April 20, 2010

ABSTRACT

Cellular responses to environmental stimuli require conserved signal transduction pathways. Inbudding yeast (Saccharomyces cerevisiae), nutrient limitation induces morphological changes that dependon the protein kinase A (PKA) pathway and the Kss1 mitogen-activated protein kinase (MAPK) pathway. Itwas unclear to what extent and at what level there is synergy between these two distinct signalingmodalities. We took a systematic genetic approach to clarify the relationship between these inputs. Weperformed comprehensive epistasis analysis of mutants lacking different combinations of all relevantpathway components. We found that these two pathways contribute additively to nutrient limitation-induced haploid invasive growth. Moreover, full derepression of either pathway rendered it individuallysufficient for invasive growth and thus, normally, both are required only because neither is maximallyactive. Furthermore, in haploids, the MAPK pathway contributes more strongly than the PKA pathway tocell elongation and adhesion, whereas nutrient limitation-induced unipolar budding is independent ofboth pathways. In contrast, in diploids, upon nutrient limitation the MAPK pathway regulates cellelongation, the PKA pathway regulates unipolar budding, and both regulate cell adhesion. Thus,although there are similarities between haploids and diploids, cell type-specific differences clearly alterthe balance of the signaling inputs required to elicit the various nutrient limitation-evoked cellularbehaviors.

EUKARYOTIC cells respond to a wide variety ofenvironmental conditions by using a compara-

tively limited number of conserved signal transductionmechanisms. Potentially, there are diverse ways inwhich such pathways can interact with and regulateeach other to integrate stimuli, generate positive ornegative feedback, and coordinate, antagonize, orpotentiate subprocesses, thereby generating by theircombinatorial action specific cellular behaviors.

External nutrient supply affects the division pattern,adhesivity, and morphology of cells and the colonies ofthe budding yeast Saccharomyces cerevisiae (Gancedo

2001; Gagiano et al. 2002; Palecek et al. 2002;Schneper et al. 2004; Truckses et al. 2004; Verstrepen

and Klis 2006; Chen and Thorner 2007; Granek andMagwene 2010). During conditions of nutrient suffi-ciency, cells are ovoid, and haploid cells bud to formdaughter cells at the cell pole corresponding to their

own birth site (an ‘‘axial’’ budding pattern), whereasdiploid cells bud alternately from their birth end andfrom the opposite cell pole (a ‘‘bipolar’’ buddingpattern). When nutrients are plentiful, the combinationof a rounder cell shape, decreased adherence, anddivision near existing cells may facilitate denser andmore rapid population of the current niche. By contrast,during conditions of nutrient limitation, cells exhibitincreased cell–cell adhesion, cell–substratum adhesion,and substratum penetration, as well as an elongatedmorphology and a ‘‘unipolar’’ pattern of proliferation,in which daughter cells arise only at the cell poleopposite their mother’s birth end. This combinationof adhesion, substratum invasion, cell elongation, anddirectional growth (division away from, as opposed tonext to, existing cells) is thought to permit a colony ofindividually non-motile yeast cells to spread out, facili-tating the exploration of its surroundings for areas ofgreater nutritional abundance.

The changes in growth characteristics elicited bynutrient limitation are termed ‘‘pseudohyphal growth’’in diploids and ‘‘invasive growth’’ in haploids or,generally, ‘‘filamentous growth’’ (FG). Pseudohyphalgrowth under nitrogen source limitation and invasive

1Present address: Department of Biochemistry, Stanford UniversitySchool of Medicine, Stanford, CA 94305-5307.

2Corresponding author: Department of Molecular and Cell Biology,Room 16, Barker Hall, University of California, Berkeley, CA 94720-3202. E-mail: [email protected]

Genetics 185: 855–870 ( July 2010)

growth under carbon source limitation involve many ofthe molecular factors and signaling pathways (for anoverview of these pathway components, please refer tothe schematic summary of the results of this study shownin Figure 7). In particular, pseudohyphal growth andinvasive growth require two classical signaling modali-ties: (i) a mitogen-activated protein kinase (MAPK)cascade and (ii) 39, 59-cyclic adenosine monophosphate(cAMP) and cAMP-dependent protein kinase A (PKA)(Lengeler et al. 2000; Pan et al. 2000; Truckses et al.2004). The MAPK cascade that promotes FG consists ofSte20 (PAK/MAPKKKK), Ste11 (MAPKKK), Ste7(MAPKK), and Kss1 (MAPK). Fus3, another MAPKactivated by Ste7 in response to a different stimulus(mating pheromone), inhibits FG (Cook et al. 1997;Madhani et al. 1997) at least in part by targeting atranscription factor (Tec1) required for FG gene ex-pression for ubiquitin- and proteasome-mediated de-struction (Bao et al. 2004; Chou et al. 2004). Of the threeisoforms of the PKA catalytic subunit encoded in the S.cerevisiae genome, Tpk2 activates filamentous growth,Tpk3 inhibits it, and the role of Tpk1 has been unclear(Robertson and Fink 1998; Pan and Heitman 1999).

One node at which the MAPK and PKA pathways areknown to converge is on the promoters of certain genes,such as FLO11 (Rupp et al. 1999), which encodes asurface glycoprotein important for cell–substratumadhesion. Tpk2-mediated phosphorylation displacesrepressor Sfl1, thereby permitting binding of transcrip-tional activator Flo8 (Pan and Heitman 2002). Simi-larly, Kss1 catalyzes phosphorylation and displacementof the related repressors Dig1 and Dig2 from trans-activator Ste12 (Cook et al. 1996; Tedford et al. 1997;Bardwell et al. 1998b), allowing its hetero-dimeriza-tion with Tec1 to induce FG genes (Cook et al. 1996;Madhani and Fink 1997; Tedford et al. 1997; Rupp et al.1999; Zeitlinger et al. 2003; Chou et al. 2006).

The MAPK and PKA pathways also share a commonupstream activator, the GTPase Ras2 (Mosch et al.1999), a homolog of mammalian H-Ras. Ras2 activatesadenylate cyclase (Toda et al. 1985), which stimulatescAMP production; cAMP binding to the regulatory(inhibitory) subunit (Bcy1) of PKA releases the activecatalytic subunits. Separately, Ras2 action stimulatesformation of the GTP-bound form of another smallGTPase, Cdc42 (Mosch et al. 1996), which both stim-ulates Ste20 and tethers nearby its immediate substrate,Ste11 (via an adaptor protein, Ste50) (Truckses et al.2006), thereby triggering the cascade that activates Kss1.

Because FG requires both MAPK and PKA activity, itseemed reasonable that these inputs might exhibitcoordinate and synergistic regulation, such that activa-tion of one pathway might promote activation of theother. Evidence obtained in some studies suggestsinstead that the MAPK and PKA pathways functionindependently, at least for regulation of diploid pseu-dohyphal growth (Pan and Heitman 1999). However,

overexpression in haploids of any of the three PKAisoforms stimulated expression of a filamentous growthreporter construct [FRE(Ty1))TlacZ], and this stimula-tion was completely dependent on Ste12 and Tec1, onwhich basis it was proposed that a component of thecAMP pathway at or downstream of the PKAs regulates acomponent of the MAPK pathway at or upstream of thetranscriptional regulators (Mosch et al. 1999). However,it has also been reported that haploid cells lackingeither Fus3 or Kss1 exhibit substantially altered cAMPlevels, suggesting regulation of the PKA pathway by theMAPK pathway (Cherkasova et al. 2003). In light ofthese ambiguities about whether the MAPK regulatesthe PKA pathway or PKA regulates the MAPK pathway orthe two function independently, we applied a compre-hensive and systematic genetic approach to analyzewhether cross-regulation between these two pathwaysis essential for nutrient limitation-evoked phenotypicresponses.

MATERIALS AND METHODS

Yeast strains and growth conditions: S. cerevisiae strains usedin this study are listed in Table 1. In the strains generated forthis study, deletion alleles represent precise deletions of thecorresponding open reading frames, except where otherwisenoted. dig1D represents the same genomic deletion as that ofthe dig1-D1THIS3 allele in JCY501; kss1D represents the samegenomic deletion as that of the kss1DThisG allele in JCY110;ste12D represents the same genomic deletion as that of theste12DTLEU2 allele in JCY512. The disruption markersKanMX4, HIS3MX6, CgHIS3, CgLEU2, CgTRP1, KlURA3,hphNT1, and natNT2 were amplified from pFA6a-KanMX4(Wach et al. 1994), pFA6a-HIS3MX6 (Wach et al. 1997),pCgH, pCgL, pCgW, pKlU, pRS306H, and pRS306N (Taxis

and Knop 2006), respectively. Unless otherwise indicated,strains were cultivated at 30� in standard rich (YP) or defined(SC) media (Burke et al. 2000) containing 2% glucose (Glc)/dextrose (D). When necessary, selection for plasmids wasmaintained by the addition of G418 (0.2 g/liter) or theomission of appropriate nutrients. Standard yeast genetictechniques were according to Burke et al. (2000).

Plasmids and recombinant DNA methods: Plasmids (Table2) were constructed and propagated in Escherichia coli usingstandard recombinant DNA methods. pRC136, pRC137, andpRC138 were generated by site-directed mutagenesis of YCpU-KSS1 and were verified by nucleotide sequence analysis.

Random spore analysis: Strains were sporulated at 26� in0.03 m potassium acetate, 0.02% raffinose. Unsporulateddiploids were eliminated by adding 2 vol of ethyl ether andvortexing for 30 sec. After a 20-min room temperatureincubation, the aqueous phase was plated onto YPD. Coloniesarising were genotyped by their auxotrophies and abilities toinduce growth arrest halos on lawns of MATa or MATacells. Inparallel, the colonies were grown on YPD plates and assessedfor invasive growth.

Invasive growth assay: Cells were grown on a solid mediumand assessed for the extent to which they remained adhered tothe plate following exposure to a gentle stream of tap water.For the experiments summarized in Table 3, random sporeanalysis was conducted for .800 spore colonies resulting fromthe sporulation of strains RCY9041, RCY9043, RCY9103, andRCY9107. The invasive growth of each colony was scored blind

856 R. E. Chen and J. Thorner

TABLE 1

Yeast strains used in this study

Strain Genotype Source or reference

JCY100 MATa his3DThisG leu2DThisG trp1DThisG ura3-52 Cook et al. (1997)JCY101 MATa otherwise isogenic to JCY100 Bardwell et al. (1998b)JCY102 MATa/MATa (JCY100 3 JCY101) Bardwell et al. (1998b)JCY107 JCY100 ste7DTura3 Cook et al. (1997)JCY110 JCY100 kss1DThisG Cook et al. (1997)JCY117 JCY100 kss1DThisG ste7DTura3 Cook et al. (1997)JCY120 JCY100 fus3DTTRP1 Cook et al. (1997)JCY127 JCY100 fus3DTTRP1 ste7DTHIS3 Cook et al. (1997)JCY130 JCY100 kss1DThisG fus3DTTRP1 Cook et al. (1997)JCY137 JCY100 kss1DThisG fus3DTTRP1 ste7DTura3 Cook et al. (1997)JCY501 JCY100 dig1-D1THIS3 dig2-D1Tura3 Bardwell et al. (1998b)JCY512 JCY100 dig1-D1THIS3 dig2-D1Tura3 ste12DTLEU2 Bardwell et al. (1998b)JCY600 JCY100 ste12DTLEU2 Bardwell et al. (1998b)RCY006 JCY100 tpk1DTKanMX4 This studyRCY007 JCY100 tpk2DTKanMX4 This studyRCY008 JCY100 tpk3DTKanMX4 This studyRCY009 JCY100 kss1DThisG tpk1DTKanMX4 This studyRCY010 JCY100 kss1DThisG tpk3DTKanMX4 This studyRCY012 JCY100 kss1DThisG tpk2DTKanMX4 This studyRCY015 JCY100 kss1DThisG tpk1DTKanMX4 tpk3DTHIS3MX6 This studyRCY016 JCY100 tpk1DTKanMX4 tpk2DTHIS3MX6 This studyRCY017 JCY100 tpk1DTKanMX4 tpk3DTHIS3MX6 This studyRCY018 JCY100 tpk2DTHIS3MX6 tpk3DTKanMX4 This studyRCY020 JCY100 ste7DTura3 tpk1DTKanMX4 This studyRCY021 JCY100 ste7DTura3 tpk2DTKanMX4 This studyRCY022 JCY100 ste7DTura3 tpk3DTKanMX4 This studyRCY023 JCY100 kss1DThisG ste7DTura3 tpk1DTKanMX4 This studyRCY024 JCY100 kss1DThisG ste7DTura3 tpk2DTKanMX4 This studyRCY025 JCY100 kss1DThisG ste7DTura3 tpk3DTKanMX4 This studyRCY026 JCY100 fus3DTTRP1 ste7DTHIS3 tpk1DTKanMX4 This studyRCY027 JCY100 fus3DTTRP1 ste7DTHIS3 tpk2DTKanMX4 This studyRCY028 JCY100 fus3DTTRP1 ste7DTHIS3 tpk3DTKanMX4 This studyRCY029 JCY100 kss1DThisG fus3DTTRP1 ste7DTura3 tpk1DTKanMX4 This studyRCY030 JCY100 kss1DThisG fus3DTTRP1 ste7DTura3 tpk2DTKanMX4 This studyRCY031 JCY100 kss1DThisG fus3DTTRP1 ste7DTura3 tpk3DTKanMX4 This studyRCY033 JCY100 flo8DTHIS3MX6 sfl1DTKanMX4 This studyRCY104 JCY100 flo8DTHIS3MX6 This studyRCY108 JCY100 tec1DTHIS3MX6 This studyRCY109 JCY100 sfl1DTKanMX4 This studyRCY121 JCY100 fus3DTTRP1 tpk1DTKanMX4 This studyRCY122 JCY100 fus3DTTRP1 tpk2DTKanMX4 This studyRCY123 JCY100 fus3DTTRP1 tpk3DTKanMX4 This studyRCY131 JCY100 kss1DThisG fus3DTTRP1 tpk1DTKanMX4 This studyRCY132 JCY100 kss1DThisG fus3DTTRP1 tpk2DTKanMX4 This studyRCY133 JCY100 kss1DThisG fus3DTTRP1 tpk3DTKanMX4 This studyRCY608 JCY100 ste12DTLEU2 tec1DTHIS3MX6 This studyRCY609 JCY100 sfl1DTHIS3MX6 ste12DTLEU2 This studyRCY640 JCY100 ste12DTLEU2 tpk1DTKanMX4 This studyRCY660 JCY100 ste12DTLEU2 tpk3DTKanMX4 This studyRCY664 JCY100 flo8DTHIS3MX6 ste12DTLEU2 tpk3DTKanMX4 This studyRCY668 JCY100 ste12DTLEU2 tec1DTHIS3MX6 tpk3DTKanMX4 This studyRCY698 JCY100 sfl1DTKanMX4 ste12DTLEU2 tec1DTHIS3MX6 This studyRCY9041 JCY102 FUS3/fus3DTKlURA3 KSS1/kss1DTCgTRP1 TPK1/tpk1DT

KanMX4 TPK2/tpk2DTHIS3MX6 TPK3/tpk3DTCgLEU2This study

RCY9043 JCY102 DIG1/dig1DTCgLEU2 DIG2/dig2DTKlURA3 KSS1/kss1DTCgTRP1 SFL1/sfl1DTKanMX4 TPK2/tpk2DTHIS3MX6

This study

RCY9103 JCY102 FLO8/flo8DTCgTRP1 FUS3/fus3DTKlURA3 STE12/ste12DTCgHIS3 TPK1/tpk1DTKanMX4 TPK3/tpk3DTCgLEU2

This study

(continued )

MAPK and PKA Contributions to Yeast Nutrient Response 857

to its genotype, which was determined in parallel. In thisexperiment, the phenotypic range permitted a graded scoringof the degree of invasive growth, as described in the legend toTable 3, which lists the average invasive growth of all coloniesof each indicated genotype. No significant differences ininvasive growth were observed between otherwise isogenicMATa and MATa colonies.

Flocculation assay: Liquid cultures were thoroughly mixed,transferred to spectrometry cuvettes, thoroughly mixed again,and immediately checked for absorbance at 600 nm (time 0).The cuvettes were then allowed to sit without agitation, andA600 values were read at various times up to 4 hr later. For allstrains, the A600 values exhibited exponential decay over time,and the times required for A600 values to halve [t0.5 (A600)]were calculated on the basis of fitting to a first-orderexponential decay curve.

Preparation of cell extracts and immunoblotting: Cellpellets were resuspended in buffer containing 25 mm Tris–Cl(pH 7.4), 150 mm NaCl, 10% glycerol, 1 mm DTT, 2 mm EDTA,5% SDS, 0.1% Triton X-100, protease inhibitors (RocheComplete EDTA-free protease inhibitor tablets), and phospha-tase inhibitors (25 mm NaF, 0.25 mm NaVO3, 0.25 mm Na3VO4,1 mm sodium pyrophosphate, 25 mm b-glycerol phosphate).Glass beads were added and samples were vortexed, boiled, andclarified by centrifugation. Samples were resolved by SDS–PAGE, transferred to nitrocellulose filters, incubated withappropriate primary antibodies—i.e., anti-Fus3 (sc-6773, SantaCruz Biotechnology), anti-Kss1 (sc-6775, Santa Cruz Biotech-nology), or anti-phospho-MAPK (#9101, Cell Signaling Tech-nology #9101)—and then with appropriate infrareddye-conjugated secondary antibodies, and visualized by usingan infrared imaging system (Odyssey; Li-Cor).

RESULTS

A PKA consensus sequence in Kss1 is not requiredfor its function in invasive growth: The optimalconsensus motif for phosphorylation by PKA is wellcharacterized and consists of two basic residues (mostoften Arg) at the �3 and �2 positions relative to the Ser(or Thr) phosphoacceptor residue, followed at 11by a hydrophobic residue: -(R/K)(R/K)3(S/T)Hpo-(Pearson and Kemp 1991). We noted that Kss1 containssuch a motif (-Lys-Arg-Ile-Ser303-Ala-) and considered thepossibility that PKA could influence Kss1 activity and/oraffect its interaction with binding partners and/orsubstrates by modifying this site. If so, then mutatingSer303 should disrupt coordination between PKA andMAPK function in FG responses.

Consequently, we generated both a nonphosphorylat-able (S303A) and two phospho-mimetic (S303D andS303E) alleles and tested their ability to support a di-agnostic hallmark of haploid FG. In wild-type cells, Kss1activity is required to elicit invasive growth (Cook et al.1997); a kss1D strain carrying an empty vector or the sameplasmid expressing either an unactivatable [Kss1(T183AY185F)] or a catalytically inactive [Kss1(K42R Q45P)]mutant cannot invade the agar, whereas invasiveness isrestored when the plasmid expresses wild-type Kss1 (Figure1, top rows). Kss1 in its inactive state (e.g., in the absence of

TABLE 1

(Continued)

Strain Genotype Source or reference

RCY9107 JCY102 DIG1/dig1DTCgLEU2 DIG2/dig2DTKlURA3 FLO8/flo8DTCgTRP1 SFL1/sfl1DTKanMX4 STE12/ste12DTCgHIS3

This study

RCY9113 JCY100 kss1DTCgTRP1 This studyRCY9123 JCY100 fus3DTKlURA3 kss1DTCgTRP1 This studyRCY9140 JCY100 dig1DTCgLEU2 dig2DTKlURA3 This studyRCY9150 JCY100 tpk3DTCgLEU2 This studyRCY9153 JCY100 ste12DTCgHIS3 This studyRCY9155 JCY100 ste12DTCgHIS3 tpk3DTCgLEU2 This studyRCY9161 JCY100 flo8DTCgTRP1 This studyRCY9163 JCY100 flo8DTCgTRP1 tpk3DTCgLEU2 This studyRCY9169 JCY100 flo8DTCgTRP1 ste12DTCgHIS3 This studyRCY9171 JCY100 flo8DTCgTRP1 ste12DTCgHIS3 tpk3DTCgLEU2 This studyRCY9179 JCY100 sfl1DTKanMX4 This studyRCY9182 JCY100 dig1DTCgLEU2 dig2DTKlURA3 sfl1DTKanMX4 This studyRCY9184 JCY100 dig1DTCgLEU2 dig2DTKlURA3 ste12DTCgHIS3 This studyRCY9185 JCY100 sfl1DTKanMX4 ste12DTCgHIS3 This studyRCY9188 JCY100 dig1DTCgLEU2 dig2DTKlURA3 sfl1DTKanMX4 ste12DTCgHIS3 This studyRCY9191 JCY100 dig1DTCgLEU2 dig2DTKlURA3 flo8DTCgTRP1 This studyRCY9192 JCY100 flo8DTCgTRP1 sfl1DTKanMX4 This studyRCY9195 JCY100 dig1DTCgLEU2 dig2DTKlURA3 flo8DTCgTRP1 sfl1DTKanMX4 This studyRCY9198 JCY100 dig1DTCgLEU2 dig2DTKlURA3 flo8DTCgTRP1 ste12DTCgHIS3 This studyRCY9199 JCY100 flo8DTCgTRP1 sfl1DTKanMX4 ste12DTCgHIS3 This studyRCY9202 JCY100 dig1DTCgLEU2 dig2DTKlURA3 flo8DT

CgTRP1 sfl1DTKanMX4 ste12DTCgHIS3This study

RCY9327 JCY102 dig1DTCgLEU2/dig1DTCgLEU2 DIG2/dig2DTKlURA3 KSS1/kss1DTCgTRP1 STE12/ste12DTCgHIS3

This study


its activator Ste7) represses invasive growth (Cook et al.1997) by remaining bound to Ste12 (Bardwell et al.1998a); a kss1D ste7D double mutant is able to invade agar,and introduction of a Kss1-expressing plasmid, but not theempty vector, abolishes the invasiveness of the kss1D ste7D

cells (Figure 1, bottom rows). In either context, theKss1(S303A), Kss1(S303D), and Kss1(S303E) mutants allbehaved indistinguishably from wild-type Kss1. Thus,elimination of Ser303 does not affect either the stimulatoryor the inhibitory functions of Kss1 in invasive growth.Therefore, modification of this residue by PKA (or anyother protein kinase) is unlikely to be of any importance formodulating Kss1 function in haploid FG.

Expression and activation of MAPKs Kss1 and Fus3are independent of PKA: Although Kss1 MAPK doesnot appear to be a direct target of PKA, it is possible thatPKA action potentiates the activity of this MAPK via anindirect mechanism. For example, in lymphocytes, PKAphosphorylates a MAPK-inactivating phosphatase, caus-ing a reduction in its binding affinity for the MAPK,thereby extending the duration of MAPK activation(Saxena et al. 1999). Among the known MAPK-targetedphosphatases in yeast, Ptp3 contains two consensus PKAmotifs (-KRAT75A- and -RKNT675M-), raising the possi-bility that PKA action might potentiate Kss1 function inFG by preventing its efficient Ptp3-mediated dephos-phorylation. Furthermore, it is claimed that, in cerebel-lar granule cells, elevated cAMP promotes (via PKA) theactivation of Ras, thereby stimulating the MAPKKK(Raf) of the cascade that activates the mammalianKss1 ortholog (Erk2) (Obara et al. 2007). To testwhether PKA activity affects Kss1 activation in any way,we assessed the extent of MAPK phosphorylation inextracts from cells lacking Tpk1, Tpk2, or Tpk3 (orcombinations of these gene products) (Figure 2). Onlymodest effects on the level of activated (dually phos-phorylated) Kss1 were observed, and they varied some-what from trial to trial. Therefore, we conclude that theabsence of any single PKA isoform or even pairs of PKAisoforms [cells lacking all three are inviable (Toda et al.1987)] does not reproducibly alter Kss1 activationunder FG conditions. Likewise, Fus3 (which in wild-type cells is unphosphorylated in the absence ofpheromone stimulation) did not become detectably

phosphorylated in any of the PKA mutant strains.Furthermore, the total levels of Kss1 and Fus3 wereunaffected by loss of PKA (Figure 2). Thus, PKA doesnot regulate the amount or activation state of Kss1 orFus3.

The function of PKA during invasive growth doesnot require Kss1 MAPK: The preceding results in-dicated that Kss1 MAPK is not regulated by PKA, eitherdirectly as a substrate or indirectly in terms of its ex-pression or activation level. Therefore, we investigatedmore generally whether any functional interrelationshipsbetween these two kinases exist that contribute to theirmutual regulation of the biological process of invasivegrowth.

Of the MAPKs involved in invasive growth, Kss1 is apositive regulator because mutation of Kss1 reduces theinvasiveness of wild-type cells, and Fus3 is a negativeregulator because mutation of Fus3 restores invasive-ness to kss1D cells (Cook et al. 1997). Among the PKAisoforms, Tpk2 is a positive regulator because tpk2D cellsare non-invasive, and Tpk3 is a negative regulatorbecause tpk3 cells are hyper-invasive (consistent withthis, tpk2/tpk2 diploids are defective for pseudohyphalgrowth and tpk3/tpk3 diploids display enhanced pseu-dohyphal growth) (Robertson and Fink 1998). Onegroup reported that tpk1D haploids and tpk1/tpk1diploids displayed indistinguishable invasive growthand pseudohyphal growth, respectively, relative towild-type cells, suggesting that Tpk1 has no role in theseprocesses (Robertson and Fink 1998); on the otherhand, another group reported that tpk1/tpk1 diploidsexhibited enhanced pseudohyphal growth, suggestingthat Tpk1 is a negative regulator of FG, at least indiploids (tpk1 haploids were not tested) (Pan andHeitman 1999). This ambiguity about the role ofTpk1 may stem from the fact that the S1278b back-grounds used in the two studies cited derive fromdifferent laboratories and have undergone differentmanipulations and, thus, are likely not strictly isogenic.In any event, Kss1 and Tpk2 positively regulate haploidinvasive growth, Fus3 and Tpk3 negatively regulate thisphenotype, and it was unclear whether Tpk1 is eitherneutral or, perhaps, a negative regulator of this behaviorin haploids.

TABLE 2

Plasmids used in this study

Plasmid Description Source or reference

pRC136 YCplac33 PKSS1-kss1 (S303A) This studypRC137 YCplac33 PKSS1-kss1 (S303D) This studypRC138 YCplac33 PKSS1-kss1 (S303E) This studyYCplac33 CEN URA3 Gietz and Sugino (1988)YCpU-KSS1 YCplac33 PKSS1-KSS1 Bardwell et al. (1998a)YCpU-kss1(K42R Q45P) YCplac33 PKSS1-kss1 (K42R Q45P) Cook et al. (1997)YCpU-kss1(T183A Y185F) YCplac33 PKSS1-kss1 (T183A Y185F) Cook et al. (1997)


Previous genetic analyses of invasive growth havegenerally been performed either on MAPK pathwaycomponents only or on PKA pathway componentsonly. To examine interpathway regulatory relation-ships, we combined tpk1, tpk2, and tpk3 null mutationswith kss1 and/or fus3 null mutations. Interestingly, likeremoval of Fus3, absence of either Tpk1 or Tpk3restored invasiveness to a kss1D strain (Figure 3).Similarly, absence of Fus3 substantially increased theinvasiveness of cells lacking Tpk2 (Figure 3). Theseresults demonstrate, first, that Tpk1 indeed performs anegative function during invasive growth, corroborat-ing the conclusions reached for its role in diploidpseudohyphal growth (Pan and Heitman 1999).Second, our findings indicate that the stimulatoryroles of Kss1 and Tpk2 are weaker than (or fartherupstream) the inhibitory roles of Fus3, Tpk1, andTpk3. Third, these data show that the requirement forKss1 can be bypassed by removal of any of severaldifferent negatively acting factors and that thesefactors contribute independently and act additively,suggesting that Fus3, Tpk1, and Tpk3 do not acttogether in a linear pathway whose inhibitory effect iscounteracted by Kss1 action. Consistent with thisconclusion, the invasiveness of the tpk2D fus3D strainwas largely lost in cells lacking Kss1 (Figure 3). Thus,even in the absence of the inhibitory factor Fus3, eitherTpk2 or Kss1 is needed in a positive sense to elicit anyresponse. The implications of these results will bediscussed in a larger context below.

A strain lacking Ste7 (the MAPKK that activates bothKss1 and Fus3) is non-invasive (Roberts and Fink

1994). Just as absence of Tpk1 or Tpk3 restoredinvasiveness to a kss1D strain (Figure 3), absence ofeither Tpk1 or Tpk3 also restored invasiveness to a ste7D

strain (Figure 3). Active Kss1 (as in STE7 strains) has anactivating function, whereas inactive Kss1 (as in ste7D

strains) acts as a repressor (Cook et al. 1997; Madhani

et al. 1997; Bardwell et al. 1998a,b). For this reason,elimination of Kss1 restores invasiveness to a ste7D strain(Cook et al. 1997). However, importantly, we found thatthe invasiveness of ste7D kss1D, kss1D fus3D, and ste7D

kss1D fus3D strains still required Tpk2 (Figure 3),indicating that Tpk2 performs an essential function ininvasive growth that cannot be mediated by the MAPKcascade.

Together, these results indicate that both promotionof invasive growth by the PKA isoform Tpk2 andinhibition of invasive growth by the PKA isoformsTpk1 and Tpk3 are independent of the action of theMAPKs Kss1 and Fus3. In other words, these twosignaling modules are not connected at the level oftheir effector kinases.

Systematic analysis of distal components of theMAPK and PKA pathways in invasive growth: Thekinases in both the MAPK and PKA pathways directlyinteract with, phosphorylate, and modify the activity oftranscriptional repressors and activators. Downstreamof Kss1 are Dig1 (repressor), Dig2 (repressor), andSte12 (activator) (Cook et al. 1997; Madhani et al. 1997;

TABLE 3

Invasive growth of MAPK and PKA pathway mutants

TPK2 tpk2D FLO8 flo8DGenotype KSS1 kss1D tpk2D kss1D STE12 ste12D flo8D ste12D

FUS3 TPK3 TPK1 1 � � x 1 –/x x xtpk1D 1/11 1� � x 1/11 �� x x

tpk3D 1/11 1 � x 1/11 1/11 x xtpk3D tpk1D 1/11 1 ND ND 1/11 1/11 x x

fus3D 1/11 1 � � 1/11 � –/x xfus3D tpk1D 1/11 1 � � 1/11 �/1� –/x xfus3D tpk3D 1/11 1 � � 1/11 1/11 –/x xfus3D tpk3D tpk1D 1/11 1 ND ND 1/11 1/11 – xSFL1 DIG1 DIG2 1 � � x 1 –/x x x

dig2D �/�� –/x x �/�� –/x x xdig1D 1/11 11 11 11 1/11 � � xdig1D dig2D 11 11 11 11 11 –/x 1 x

sfl1D 11 11 11 11 11 11 � �sfl1D dig2D 11 11 11 11 11 11 � –/xsfl1D dig1D 11 11 11 11 11 11 1/11 �sfl1D dig1D dig2D 11 11 11 11 11 11 11 –/x

Invasive growth was scored as follows: 11, no erosion of growth patch by water stream. 1, only slight erosion of growth patch bywater stream (defined as that displayed by wild-type strain JCY100). 1�, some significant erosion, but majority of growth patchremained. �, borderline cases not unambiguously closer to either ‘‘1�’’ or ‘‘��’’. ��, non-invasive, but after wash-off by waterstream some residual material remained. –, non-invasive, but after wash-off by water stream, patch imprint readily detectable. x,non-invasive, and after wash-off by water stream, patch imprint difficult to detect. ND, not determined; no spores of these gen-otypes were obtained because tpk1D tpk2D tpk3D mutants are inviable (Toda et al. 1987).


Tedford et al. 1997; Bardwell et al. 1998a,b). Down-stream of Tpk2 are Sfl1 (repressor) and Flo8 (activator)(Robertson and Fink 1998; Pan and Heitman 1999,2002). To determine whether any regulatory interrela-tionships exist among these transcriptional modulators,we assessed invasive growth in strains lacking combina-tions of these pathway components. Toward this end,four diploid strains were constructed. Two were hetero-zygous for the activating kinases (KSS1/kss1 TPK2/tpk2),and two of the strains were heterozygous for thetranscriptional activators (STE12/ste12 FLO8/flo8).From each strain, derivatives that were heterozygousfor either the inhibitory kinases (TPK1/tpk1 TPK3/tpk3FUS3/fus3) or the transcriptional repressors (DIG1/dig1DIG2/dig2 SFL1/sfl1) were generated. Such quintuplyheterozygous diploid strains were allowed to sporulate,generating haploid spores containing various combina-tions of PKA and MAPK pathway mutations, eachidentifiable by a distinct associated marker. Thesespores were plated and grown into colonies, andcolonies were assessed for their invasive growth pheno-type blind to its genotype, which was determined sub-sequently. Table 3 catalogs the average invasive growthphenotype of all colonies of each genotype (otherwiseisogenic MATa and MATa colonies exhibited verysimilar properties).

Typically, determination of the invasive growth phe-notype is binary—i.e., strains are scored as invasive ornon-invasive. Due to the very large number of coloniesanalyzed in parallel (.800) in our experiments, we wereable to distinguish a wider phenotypic range and assignfiner gradations to the strength of the invasive growthresponse (see legend to Table 3). We found that strainsof different genotypes reproducibly exhibited distinctdegrees of invasiveness over a rather broad range (Table3). Thus, our systematic analysis revealed that invasivegrowth is not an all-or-none response, but rather aquantitative or at least semiquantitative trait.

The MAPK and PKA pathways contribute additivelyto promote invasive growth: Cells lacking all three PKAcatalytic subunits are inviable (Toda et al. 1987). Theinviability of a tpk1D tpk2D tpk3D triple mutant can berescued, however, by removal of another protein kinase,Yak1 (Garrett and Broach 1989), suggesting that theessential function of PKA is to antagonize the action of

Yak1. Consistent with this conclusion, mass spectrome-try analysis has shown that Yak1 is phosphorylated in vivoat its five PKA consensus sites (Zappacosta et al. 2002).In our analysis (Table 3), no strains were obtained inwhich all three PKA catalytic subunits were deleted,regardless of the presence or absence of one or both ofthe MAPKs, indicating that neither Kss1 nor Fus3 has arelationship to PKA similar to that of Yak1.

Scoring of the invasive growth phenotype on a finerscale revealed some previously unrecognized interpath-way relationships. For example, Tpk1 can clearly be seento be a negative regulator of invasive growth not onlybecause its loss restores invasiveness to a non-invasivekss1D mutant (Figure 3), but also because its absenceenhanced the invasiveness of even an otherwise wild-type strain (Table 3).

More generally, our analysis could provide insightinto the interconnections among the positively actingand negatively acting factors, as follows. Suppose X andY are two signal transduction proteins whose corre-sponding mutations, denoted x and y, give rise toopposing phenotypes. If the phenotype of the double-mutant x y is identical to the phenotype of one of thesingle mutants, for example, y, then this suggests that Yacts downstream of X in a single pathway. For example,as shown above (Figure 3), because a kss1D fus3D straindisplays the same phenotype as a fus3D strain (invasive),but not a kss1D strain (non-invasive) (Cook et al. 1997),this suggests that Fus3 acts, in a genetic sense, down-stream of Kss1 in its regulation of invasive growth.Formally, this relationship means that the sole role ofKss1 is to counteract the effects of Fus3. If, however, thephenotype of x y is intermediate between those of x andy, then X and Y are likely to contribute independently tothe net phenotype.

The invasive growth behavior of strains lackingvarious combinations of the kinases is shown in Table3 (note that Table 3 includes all of the STE7 genotypesdepicted in Figure 3 and also others). The extent ofinvasiveness corresponds to genotype in a graded, asopposed to abrupt, manner. Overall, our schemepermitted us to deduce certain rankings. For example,Tpk2 is a more potent activator of invasive growth thanKss1, and Fus3 is a more potent repressor than Tpk3,which is a more potent repressor than Tpk1. In no case

Figure 1.—Mutation of a PKA consensus phos-phorylation motif in Kss1 does not affect either itspositive or its negative functions in invasive growth.StrainsJCY110(kss1D)andJCY117(kss1D ste7D)weretransformed withan empty vector(vec; YCplac33) orwith plasmids encoding Kss1 (wt; YCpU-KSS1),Kss1(K42R Q45P) [KR; YCpU-kss1(K42R Q45P)],Kss1(T183A Y185F) [TA/YF; YCpU-kss1(T183AY185F)], Kss1(S303A) (SA; pRC136), Kss1(S303D)(SD; pRC137), or Kss1(S303E) (SE; pRC138). Trans-formants were grown on SCD–Ura plates, replicatedto YPD, and assessed for invasive growth.


did we discern that loss of Tpk1 or Tpk3 caused anyincrease in the invasiveness of a strain already lackingTpk2, indicating that the inhibitory functions of Tpk1and Tpk3 are exerted on Tpk2; i.e., they likely functionin the same pathway. With the exception of theserelationships among the PKA isoforms, invasivenessincreases diagonally from the upper right corner ofthe quadrant to the lower left, because it can beenhanced either by adding another activatory kinaseor by deleting another inhibitory kinase, suggesting thatthe effects of PKA and the MAPKs Kss1 and Fus3contribute independently and additively to invasivegrowth. Indeed, combining the mutations in an invasivestrain with the mutations in a non-invasive straingenerally caused the resultant strain to exhibit interme-diate invasiveness. Importantly, all strains lacking theactivators Tpk2 and Kss1 are less invasive than theotherwise identical strains lacking only one of thesefactors; this would not be the case if one was upstream ofthe other in a single pathway. In agreement with the datapresented in earlier sections indicating that PKA iso-forms do not function upstream of MAPK isoformsduring invasive growth, the results compiled in Table 3further broaden and support the conclusion that thePKA and MAPK kinases act independently and addi-tively to facilitate the FG response.

Analysis of mutants containing combinations of theactivating kinases (Kss1 and Tpk2) and the transcrip-tional repressors (Dig1, Dig2, and Sfl1) is shown inTable 3. As previously observed, cells lacking Kss1, Dig1,and Dig2 were slow growing (Breitkreutz et al. 2003),which is most likely due to the toxicity of hyperactiveSte12 (Dolan 1996), and quickly gave rise to faster-growing variants. Indeed, tetrad analysis of adig1DTCgLEU2/dig1DTCgLEU2 DIG2/dig2DTKlURA3KSS1/kss1DTCgTRP1 STE12/ste12DT CgHIS3 strain(RCY9327) revealed that the growth defect of a kss1D

dig1D dig2D mutant was completely rescued by loss ofSte12. It has been reported that deletion of DIG2

enhances the invasiveness of a dig1D strain, reducesthe invasiveness of fus3D and fus3D kss1D strains, anddoes not affect the invasiveness of otherwise wild-typestrains (Breitkreutz et al. 2003). Consistent with theapparent dual nature of the Dig2 function, in ouranalysis, deletion of DIG2 increased the invasiveness ofa dig1D strain and decreased the invasiveness of a tpk2D

strain (Table 3). However, in our hands, loss of Dig2 inan otherwise wild-type strain detectably diminishedinvasive growth. In contrast to Dig2, the loss of Dig1(and/or Sfl1) caused hyper-invasiveness even in theabsence of both Kss1 and Tpk2 (Table 3). This func-tional difference between Dig1 and Dig2 is consistentwith reports that Dig1 and Dig2 interact with separateregions of Ste12 (Olson et al. 2000) and that Dig1 andDig2 incorporate differentially into transcription factorcomplexes (Ste12-Dig1-Dig2 and Tec1-Ste12-Dig1)(Chou et al. 2006), which is consistent with distinctmechanisms of action.

We observed that kss1D tpk2D dig1D and kss1D tpk2D

sfl1D strains are more strongly hyper-invasive than fus3D

tpk1D tpk3D strains (Table 3). These results suggest, first,that Fus3, Tpk1, and Tpk3 do not inhibit transcriptionas strongly as Dig1 and Sfl1 do. Second, this findingreveals that derepression of the transcriptional appara-tus downstream of either the MAPK or the PKA pathwayis sufficient to bypass the need for the upstreamsignaling kinases in either pathway. Thus, the contribu-tions of the two pathways are not specialized with respectto invasive growth, and the sole essential reason for thecombined function of Kss1 MAPK and PKA duringfilamentous growth is to elicit a sufficiently potent levelof transcriptional induction from a common set ofgenes. There is evidence that optimal invasive growthmay require cap-independent translation of certainmRNAs (Gilbert et al. 2007). Our observations suggesteither that cap-independent translation during invasivegrowth does not require either Kss1 or Tpk2 or that cap-independent translation becomes dispensable when

Figure 2.—Absence of PKA does not affect lev-els or activation of Fus3 or Kss1. Whole-cell ex-tracts from exponentially growing cultures ofstrains JCY100, JCY110, JCY120, RCY006,RCY007, RCY008, RCY015, RCY016, RCY017,and RCY018 were immunoblotted to assess levelsof phospho-MAPK, Kss1, and Fus3. Strains lack-ing Kss1 or Fus3 serve to identify the appropriatebands. The bottom two images were spliced to re-order lanes for ease of comparison.


Dig1- or Sfl1-regulated transcription targets becomehighly derepressed. Additionally, it has been reportedthat glucose depletion triggers rapid PKA-dependentinhibition of translation (Ashe et al. 2000; Uesono et al.2004). Our results suggest that this acute response is nota critical component of the much longer process ofachieving efficacious invasive growth.

The phenotypes of mutants lacking the transcrip-tional activators Ste12 or Flo8, or both, combined withthe loss of the inhibitory kinases Tpk1, Tpk3, and/orFus3 or with the absence of the transcriptional repress-ors Dig1, Dig2, and Sfl1, are shown in Table 3. Strainslacking both Flo8 and Ste12 were non-invasive even inthe absence of the inhibitory kinases or the repressors,indicating that the combined effect of these negativeregulators is exerted solely via their regulation on Flo8and Ste12. Because loss of Sfl1 caused a ste12D flo8D

strain to display a slight degree of invasiveness, it ispossible that Sfl1 negatively regulates one or more FGgenes that are not Ste12 and/or Flo8 targets. Moststrikingly, however, absence of Sfl1 (the Flo8 repressor)restored invasiveness to ste12D and ste12D tec1D strains(but not to a flo8D strain) (Table 3 and Figure 4A) and,reciprocally, absence of Dig1 and Dig2 (the Ste12repressors) restored invasiveness to a flo8D strain (butnot to a ste12D strain), in further agreement with theview that the PKA and MAPK inputs act separately andadditively.

Although invasiveness could be fully restored to aflo8D strain by loss of Dig1 and Dig2 (or Dig1 and Sfl1),it was only slightly increased by loss of Fus3, Tpk1, andTpk3, suggesting that, in comparison to the repressors,inhibition of Ste12 by these kinases is rather weak. Incontrast, invasiveness was restored to ste12D and ste12D

tec1D strains not only by loss of Sfl1, but also by loss ofTpk3 (Table 3 and Figure 4B), suggesting that thisupstream kinase is, operationally, as potent an inhibitorof Flo8 activity as its directly interacting transcriptionalrepressor. Although invasiveness was restored to a kss1D

strain by removing either Tpk1 or Tpk3, invasivenesscould be restored to a ste12D strain only by removingTpk3, and not by eliminating Tpk1, which furtherhighlights that these two PKA isoforms have differentinhibitory strengths.

In summary, complete derepression of the transcrip-tional activator controlled by the MAPK pathway, or of thetranscriptional activator controlled by the PKA pathway, issufficient for invasive growth, even in the completeabsence of the other pathway. Thus, each of these twoinputs does not contribute unique activities or specializedfunctions that must be integrated with the other in somecombinatorial fashion to elicit the desired response;rather, each signal impinges separately and additively toachieve the required level of total signal strength.

Glucose limitation-induced unipolar budding pat-tern is independent of Ste12 and Flo8: In addition toagar invasion and increased cell–cell adhesion (see nextsection), haploids undergoing FG also exhibit a slightlyelongated morphology and a unipolar budding pattern.To investigate the roles of the MAPK and PKA signalingpathways in regulating these other phenotypic behaviorsassociated with FG, we analyzed the mutants generatedfor our epistasis analysis that yielded the most informativeand dramatic insights about how the PKA and MAPKpathways contribute to agar invasiveness (Table 3).

Cells lacking Ste12 were generally rounder (lessovoid) than STE12 cells when grown in a mediumcontaining excess glucose (YPD); cells lacking both

Figure 3.—In the absence of Tpk1 or Tpk3,the MAPK cascade is not necessary for invasivegrowth, and in the absence of the MAPK cascade,Tpk2 is still required for invasive growth. StrainsJCY100, JCY107, JCY110, JCY117, JCY120,JCY127, JCY130, JCY137, RCY006, RCY007,RCY008, RCY009, RCY010, RCY012, RCY020,RCY021, RCY022, RCY023, RCY024, RCY025,RCY026, RCY027, RCY028, RCY029, RCY030,RCY031, RCY121, RCY122, RCY123, RCY131,RCY132, and RCY133 were assessed for inva-sive growth.


Ste12 and Flo8 were generally rounder than STE12FLO8 cells when grown in a glucose-limiting (YP�)medium (Table 4). All of the strains examined ex-hibited axial budding in rich medium, with theexception of STE12 strains lacking both Dig1 andDig2, which exhibited unipolar budding (and moreelongated cells) under the same conditions (Table 4and Figure 5). Upon a shift to a glucose-limitingmedium, all of the strains exhibited unipolar budding(Table 4). Thus, the switch in cell division pattern thatoccurs upon glucose limitation is independent of bothSte12 and Flo8.

Ste12 is a more potent inducer of cell–cell adhesionthan Flo8: To assess cell–cell adhesion (flocculation),we measured the rate at which cells settled out of liquidcultures (Figure 6) and also examined by microscopythe extent of cell clustering in liquid cultures (Table 4).Surprisingly, glucose limitation, which is known totrigger invasive growth and unipolar budding (Cullen

and Sprague 2000), had relatively little effect onflocculation and cell clustering, at least as comparedto the markedly more pronounced effects caused bymutations (Figure 6 and Table 4). Absence of Tpk3 fromotherwise wild-type cells caused an increase in floccula-tion (shorter time of cell settling), but the most highlyflocculent strains were those lacking the transcriptionalrepressors Dig1 and Dig2, or Sfl1 (Figure 6). Interest-ingly, the hyper-flocculation of the sfl1D and tpk3D

strains required the presence of both Flo8 and Ste12,whereas the dig1D dig2D strains were hyper-flocculenteven in the absence of Flo8. The flocculation behaviorof the mutants tested correlated well with the size of thecell clusters observed by microscopy; specifically, in-

creased flocculation corresponded to larger average cellcluster size (Table 4). In toto, these results indicate that,similar to their functions in invasive growth, theadhesion-inducing functions of Flo8 and Ste12 are notmaximally activated in wild-type cells even under nutri-ent-limiting conditions. Furthermore, it is clear thatSte12 is a much more potent inducer of cell–celladhesion than Flo8, since derepression of Ste12 permitsenhanced adhesion even in the absence of Flo8,whereas adhesion is not significantly increased by de-letion of repressors in strains lacking Ste12.

DISCUSSION

The MAPK Kss1 activates the transcription factorSte12 by relieving repression imposed by Dig1 andDig2 (Bardwell et al. 1998a,b). The PKA Tpk2 activatesthe transcription factor Flo8 by relieving repressionimposed by Sfl1 (Robertson and Fink 1998; Pan andHeitman 2002). Both of these kinase-regulated tran-scription circuits are necessary for nitrogen limitation-induced pseudohyphal growth in diploids and glucoselimitation-induced invasive growth in haploids. Thesenutrient limitation-induced responses are accompaniedby cell–cell and cell–substratum adhesion, cell elonga-tion, and a unipolar budding pattern. Here we analyzedthe contributions of the MAPK and PKA pathways tothese behaviors by using a comprehensive and system-atic genetic approach and examined the extent to whichagar invasion is representative of the other FG-associatedphenotypes in haploids. A summary of our conclusions isshown in schematic form in Figure 7.

Figure 4.—Lack of the PKA isoform Tpk3 orthe transcriptional repressor Sfl1 renders thetranscriptional activator Ste12 unnecessary forinvasive growth. (A) Strains JCY100, JCY600,RCY033, RCY104, RCY109, RCY608, RCY609,and RCY698 were assessed for invasive growth.(B) Strains JCY100, JCY600, RCY006, RCY008,RCY104, RCY108, RCY640, RCY660, RCY664,and RCY668 were assessed for invasive growth.


Functions of the MAPK and PKA pathways forcellular response to nutrient limitation: It has beenobserved that, in diploids growing on nitrogen-limitedmedia, Ste12 is necessary for cell elongation and normal(linear) cell–cell junctions, whereas Tpk2 is necessary fornormal cell–cell junctions and the switch from bipolar tounipolar budding (Pan and Heitman 1999). Thus, it wasconcluded that, in diploids, the MAPK pathway specifi-cally regulates cell elongation, the PKA pathway specifi-cally regulates unipolar budding, and both pathwaysregulate cell adhesion (Pan and Heitman 1999).

As we observed in this study, in haploids, Ste12regulates cell elongation, but Flo8 also contributesduring glucose limitation (Table 4). However, thesedifferences were rather subtle because the nutrientlimitation-induced elongation of haploid cells is muchless pronounced than in diploids (Galitski et al. 1999).Cell–cell adhesion was relatively independent of glucoselevels, but much more strongly dependent on Ste12than on Flo8 (Figure 6 and Table 4). The switch fromaxial to unipolar budding upon glucose limitation in

haploids was, by contrast, independent of both Ste12and Flo8 (Table 4 and Figure 5), as well as independentof Tpk2 (Chen 2008). It is possible that this Ste12- andFlo8-independent effect may be mediated by the actionof a third class of protein kinase, Snf1 (AMP-activatedprotein kinase), which becomes activated upon glucoselimitation (Hardie et al. 1998; Hedbacker and Carlson

2008) and contributes to FG by antagonizing two zinc-finger repressors, Nrg1 and Nrg2 (Vyas et al. 2001, 2003;Kuchin et al. 2002). Thus, in haploids, cell elongationand especially cell–cell adhesion are primarily regulatedby the MAPK pathway, with a more limited contributionby the PKA pathway, whereas glucose limitation-inducedunipolar budding is independent of both of thesepathways, which is somewhat different from the regula-tory hierarchy required for eliciting similar responses indiploids. Indeed, it has been noted that expression ofFlo11, a cell-surface glycoprotein required for FG inboth haploids and diploids, is �103 higher in haploidsthan in diploids (Mosch et al. 1999). Taken togetherwith our results, it is clear that, despite the similarities in

Figure 5.—Glucose limitation-induced unipo-lar budding pattern is independent of Ste12 andFlo8. Wild type (WT; JCY100), flo8D (RCY104),ste12D (JCY600), and dig1D dig2D (JCY501) cellswere examined by microscopy. To facilitate as-sessment of budding patterns, cell walls and celldivision junctions were stained with calcofluorwhite (0.1 g/liter) just before visualization.YPD: cultures were grown to exponential phasein YPD. YP�: overnight YPD cultures were diluted103 directly into YP and incubated for 16 hr.


cellular behaviors that accompany FG in diploids andhaploids, and despite the fact that the same kinasepathways are involved, cell type-specific differencesimpose different regulatory dependencies and out-comes in the face of similar inputs. It will be interestingto compare the detailed regulatory arrangements inother yeasts and pathogenic fungi, whose filamentousbehaviors generally require components orthologous tothe MAPK and PKA signaling pathways (and the down-stream transcription factors) first described in S. cerevisiae(Bruno et al. 1996; Banuett 1998; Martinez-Espinoza

et al. 2004; Verstrepen and Klis 2006; Klosterman et al.2007; Rispail et al. 2009; Hoi and Dumas 2010).

Subthreshold signaling through the MAPK and PKApathways combines additively to promote agar invasion:Prior to our study, the existence and direction of cross-regulation between the Kss1 MAPK and PKA pathways inthis developmental process have been unclear. Someevidence indicated that the MAPK and PKA pathwaysfunction independently, at least in the regulation ofdiploid pseudohyphal growth (Pan and Heitman

1999), whereas other findings suggested that, forhaploid invasive growth, the PKA pathway regulatesthe MAPK pathway (Mosch et al. 1999). However,another study concluded the converse, namely thatthe MAPK pathway regulates the PKA pathway inhaploids (Cherkasova et al. 2003).

As documented here, to clarify the relationshipbetween the MAPK and PKA pathway inputs in haploidFG, we performed a systematic genetic analysis usingcross-combinations of pathway mutants. Each of thesepathways contains multiple terminal kinase isoforms.Among the overarching conclusions of this work arethat certain isoforms are positive regulators and othersare negative regulators and, within each class, regulatorsthat exert control in the same direction nonetheless doso with distinctly different strength. Most revealing, we

found that positively acting kinases are individuallynecessary, but only to relieve the repression imposedby the negatively acting kinases. Furthermore, eachsignaling pathway regulates a discrete transcriptionalactivation apparatus, including distinct repressors thatact on different activators. Both transcriptional activa-tors of each pathway are normally required for haploidinvasive growth, but only because neither activator ismaximally active, even under nutrient-limiting condi-tions, and thus input from both signaling pathways isrequired to reach an adequate level of expression toelicit the appropriate cellular behavior (Figure 7). Inother words, the contributions of the Kss1 MAPKpathway and the PKA pathway are additive and over-lapping (nearly redundant) in function because fullderepression of either transcriptional activator is suffi-cient for invasive growth, even in the total absence ofinput from the other pathway.

Differential control of agar invasion, flocculence,cell elongation, and unipolar budding: We found thatagar invasion, cell elongation, and flocculation arestrongly dependent on Ste12. Nonetheless, we were ableto distinguish different degrees of dependence of floc-culation, cell elongation, unipolar budding, and agarinvasion on the MAPK pathway, the PKA pathway, andglucose limitation. Thus, in a certain sense, these pro-cesses are not tightly coupled and are to some extentseparable. For example, strains that exhibited the mostrobust agar invasion were not uniformly those thatwere the most flocculent, and the glucose withdrawal-induced switch to a unipolar budding pattern was notaccompanied by a notable increase in flocculence. Thus,although both sfl1D ste12D (Flo8 on, Ste12 off) and dig1D

dig2D flo8D (Flo8 off, Ste12 on) strains display agarinvasion, they are not identical. In any event, the re-quirement in wild-type cells for both the MAPK and PKAinputs for optimal haploid FG is to achieve sufficient

TABLE 4

Cell morphology, cell–cell adhesion, and budding pattern of MAPK and PKA pathway mutants

Genotype 1 flo8D ste12D flo8 ste12D

1

tpk3D A1, a1sfl1D A1, a1; b A1dig1D dig2D A2, a1; B, b A2, a1; B, bdig1D dig2D sfl1D A2, a2; B, b A2, a2; B, b a1

Cultures of strains JCY100, RCY9140, RCY9150, RCY9153, RCY9155, RCY9161, RCY9163, RCY9169, RCY9171,RCY9179, RCY9182, RCY9184, RCY9185, RCY9188, RCY9191, RCY9192, RCY9195, RCY9198, RCY9199, andRCY9202 were examined by microscopy. YPD: cultures were grown to exponential phase in YPD. YP�: overnightYPD cultures were diluted 103 directly into YP and incubated for 16 hr. Unless otherwise indicated, cell clusterswere small in both YPD and YP�, and cells exhibited axial budding in YPD and unipolar budding in YP�. ste12Dstrains tended to be rounder (less ovoid) than STE12 strains in YPD. ste12D, flo8D and ste12D flo8D strainstended to be rounder than STE12 FLO8 strains in YP�. A1: medium-sized clusters in YPD; a1: medium-sizedclusters in YP�; A2: very large, tight cell clusters in YPD; a2: very large, tight cell clusters in YP�; B: unipolarbudding and markedly elongated cells in YPD; b: elongated cells in YP�.


signal strength, not to yield the sum of any specificsubprocesses uniquely evoked by either pathway.

Physiological rationale for combining multiple par-tially derepressed pathways: Although both Ste12 andFlo8 are necessary for optimal invasive growth, ourresults suggest that neither is maximally active. Whatadvantages might there be in using two partially activepathways instead of one fully active one?

A common interpretation for such a situation is thatrequiring two pathways permits the integration ofupstream signals. The extent to which that occurs inthis particular system is currently unclear, as it has notbeen shown that haploid invasive growth requiressimultaneous stimulation by multiple independentlysensed stimuli. In any case, the existence of a sharedfactor, Ras2, upstream of both the MAPK and PKApathways (Mosch et al. 1999) suggests that signalintegration could in principle occur at that nodewithout the necessity for rebranching downstream ofRas2 and reintegrating at target gene promoters.

A different advantage of using two partially activepathways is that it avoids strong activation of eitherpathway, which may lead to negative side effects due toprocesses for which there are unique consequences. For

example, as revealed by the properties of dig1D dig2D

cells, partially unregulated Ste12 activity leads to cellelongation, greater cell clumping, grossly misshapenmorphology, and extraordinary flocculence, which maybe quite deleterious in the wild. Removal of all of thenegative controls known to impinge directly on Ste12(kss1D dig1D dig2D cells) resulted in a pronouncedgrowth defect that was completely rescued by alsoeliminating Ste12. Thus, full derepression of Ste12 inresponse to any signal would presumably negativelyimpact cell proliferation. In fact, the need to holdSte12 in check may explain why it is subject to at leastthree different negative regulators and cannot be fullyderepressed under nutrient-limiting conditions if thegrowth needed for FG is going to ensue, which takesplace over the course of many hours, and even days.Indeed, and in contrast, upon exposure to matingpheromone, haploid cells undergo an acute cell cyclearrest on the 20- to 30-min time scale that is due, in part,to Ste12-dependent gene regulation (Dolan 1996).Thus, costimulatory input from the PKA pathway actingvia Flo8 avoids the necessity of full-bore activation ofSte12 via the MAPK pathway. This use of partial activationvia two distinct pathways with a shared function during

Figure 6.—Ste12 is a more potent activator offlocculation than Flo8. Strains of the indicatedgenotypes were assessed for flocculation. Circleareas depict the strength of flocculation on thebasis of the rate of settling: the larger the circle,the stronger the flocculation. YPD (blue): cul-tures were grown to exponential phase in YPD;YP� (red): overnight YPD cultures were diluted103 directly into YP and incubated for 16 hr.Strains used were JCY100, RCY9123, RCY9140,RCY9150, RCY9153, RCY9155, RCY9161,RCY9163, RCY9169, RCY9171, RCY9179,RCY9182, RCY9184, RCY9185, RCY9188,RCY9191, RCY9192, RCY9195, RCY9198,RCY9199, and RCY9202.


the regulation of haploid invasive growth may reflect thenecessity to avoid hyper-activation of certain pathway-specific side processes that may be counterproductive tothe appropriate biological outcome.

If two discrete, largely parallel, signaling routes areactivated by the same stimulus, but under somewhatdifferent threshold conditions (in this case, at differentlevels as the glucose concentration falls), anotherpotential advantage of such synergy is that it allows fora reasonable degree of homeostasis in biological re-sponse as the conditions change. Yet another potentialadvantage of controlling cell behavior via two parallelinputs—each operating at an intermediate level ratherthan via one pathway functioning at its maximumlevel—is that it may help the cell avoid irreversiblecommitment to a fate from which it may not be able torecover easily. For example, the ability to rapidly switchfrom filamentous to yeast-form growth and back, and/or from invasive to non-invasive growth and back, mightprovide distinct advantages to the cell in competition

with other species or mutants that cannot so readilyadapt. All of these regulatory strategies may also occur inthe control of other developmental processes in morecomplex organisms.

We thank all members of the Thorner laboratory for advice andconstructive suggestions and especially Michael McMurray andFrancxoise Roelants for their insights and critical reading of themanuscript. This work was supported by National Institutes of Health(NIH) Predoctoral Traineeship GM07232 (to R.E.C.) and by NIHResearch GM21841 (to J.T.).

LITERATURE CITED

Ashe, M. P., S. K. De Long and A. B. Sachs, 2000 Glucose depletionrapidly inhibits translation initiation in yeast. Mol. Biol. Cell 11:833–848.

Banuett, F., 1998 Signalling in the yeasts: an informational cascadewith links to the filamentous fungi. Microbiol. Mol. Biol. Rev. 62:249–274.

Bao, M. Z., M. A. Schwartz, G. T. Cantin, J. R. Yates, III, and H. D.Madhani, 2004 Pheromone-dependent destruction of theTec1 transcription factor is required for MAP kinase signalingspecificity in yeast. Cell 119: 991–1000.

Figure 7.—Signaling circuitry involved in theresponse of haploid cells to nutrient limitation.This study focused on the relative contributionsof distal components in these pathways that haveeither positive (green) or negative (red) roles.Arrows, activation; T-bars, inhibition, inactiva-tion, or repression. Line thickness indicates thestrength of the observed effect: the thicker theline, the stronger the effect. The various effectsdepicted contribute in an additive (not all-or-none) fashion.


Bardwell, L., J. G. Cook, D. Voora, D. M. Baggott, A. R. Martinez

et al., 1998a Repression of yeast Ste12 transcription factor by di-rect binding of unphosphorylated Kss1 MAPK and its regulationby the Ste7 MEK. Genes Dev. 12: 2887–2898.

Bardwell, L., J. G. Cook, J. X. Zhu-Shimoni, D. Voora and J.Thorner, 1998b Differential regulation of transcription: re-pression by unactivated mitogen-activated protein kinase Kss1 re-quires the Dig1 and Dig2 proteins. Proc. Natl. Acad. Sci. USA 95:15400–15405.

Breitkreutz, A., L. Boucher, B. J. Breitkreutz, M. Sultan,I. Jurisica et al., 2003 Phenotypic and transcriptional plasticitydirected by a yeast mitogen-activated protein kinase network. Ge-netics 165: 997–1015.

Bruno, K. S., R. Aramayo, P. F. Minke, R. L. Metzenberg andM. Plamann, 1996 Loss of growth polarity and mislocalizationof septa in a Neurospora mutant altered in the regulatory subunitof cAMP-dependent protein kinase. EMBO J. 15: 5772–5782.

Burke, D., D. Dawson and T. Stearns, 2000 Methods in Yeast Genet-ics: A Cold Spring Harbor Laboratory Course Manual. Cold SpringHarbor Laboratory Press, Cold Spring Harbor, NY.

Chen, R. E., 2008 Function and regulation of mitogen-activated pro-tein kinases in the yeast Saccharomyces cerevisiae. Ph.D. Thesis, Uni-versity of California, Berkeley.

Chen, R. E., and J. Thorner, 2007 Function and regulation inMAPK signaling pathways: lessons learned from the yeast Sac-charomyces cerevisiae. Biochim. Biophys. Acta 1773: 1311–1340.

Cherkasova, V. A., R. McCully, Y. Wang, A. Hinnebusch andE. A. Elion, 2003 A novel functional link between MAP kinasecascades and the Ras/cAMP pathway that regulates survival. Curr.Biol. 13: 1220–1226.

Chou, S., L. Huang and H. Liu, 2004 Fus3-regulated Tec1 degrada-tion through SCFCdc4 determines MAPK signaling specificityduring mating in yeast. Cell 119: 981–990.

Chou, S., S. Lane and H. Liu, 2006 Regulation of mating and fila-mentation genes by two distinct Ste12 complexes in Saccharomy-ces cerevisiae. Mol. Cell. Biol. 26: 4794–4805.

Cook, J. G., L. Bardwell, S. J. Kron and J. Thorner, 1996 Twonovel targets of the MAP kinase Kss1 are negative regulators ofinvasive growth in the yeast Saccharomyces cerevisiae. GenesDev. 10: 2831–2848.

Cook, J. G., L. Bardwell and J. Thorner, 1997 Inhibitory and ac-tivating functions for MAPK Kss1 in the S. cerevisiae filamentous-growth signalling pathway. Nature 390: 85–88.

Cullen, P. J., and G. F. Sprague, Jr., 2000 Glucose depletion causeshaploid invasive growth in yeast. Proc. Natl. Acad. Sci. USA 97:13619–13624.

Dolan, J. W., 1996 Novel aspects of pheromone-induced cell-cyclearrest in yeast. Curr. Genet. 30: 469–475.

Gagiano, M., F. F. Bauer and I. S. Pretorius, 2002 The sensing ofnutritional status and the relationship to filamentous growth inSaccharomyces cerevisiae. FEMS Yeast Res. 2: 433–470.

Galitski, T., A. J. Saldanha, C. A. Styles, E. S. Lander and G. R.Fink, 1999 Ploidy regulation of gene expression. Science285: 251–254.

Gancedo, J. M., 2001 Control of pseudohyphae formation in Sac-charomyces cerevisiae. FEMS Microbiol. Rev. 25: 107–123.

Garrett, S., and J. Broach, 1989 Loss of Ras activity in Saccharo-myces cerevisiae is suppressed by disruptions of a new kinasegene, YAKI, whose product may act downstream of the cAMP-dependent protein kinase. Genes Dev. 3: 1336–1348.

Gietz, R. D., and A. Sugino, 1988 New yeast-Escherichia coli shuttlevectors constructed with in vitro mutagenized yeast genes lackingsix-base pair restriction sites. Gene 74: 527–534.

Gilbert, W. V., K. Zhou, T. K. Butler and J. A. Doudna, 2007 Cap-independent translation is required for starvation-induced differ-entiation in yeast. Science 317: 1224–1227.

Granek, J. A., and P. M. Magwene, 2010 Environmental and ge-netic determinants of colony morphology in yeast. PLoS Genet.6: e1000823.

Hardie, D. G., D. Carling and M. Carlson, 1998 The AMP-activated/SNF1 protein kinase subfamily: Metabolic sensors ofthe eukaryotic cell? Annu. Rev. Biochem. 67: 821–855.

Hedbacker, K., and M. Carlson, 2008 SNF1/AMPK pathways inyeast. Front. Biosci. 13: 2408–2420.

Hoi, J. W., and B. Dumas, 2010 STE12 and STE12-like proteins: fun-gal transcription factors regulating development and pathoge-nicity. Eukaryot. Cell 9: 480–485.

Klosterman, S. J., M. H. Perlin, M. Garcia-Pedrajas, S. F. Covert

and S. E. Gold, 2007 Genetics of morphogenesis and patho-genic development of Ustilago maydis. Adv. Genet. 57: 1–47.

Kuchin, S., V. K. Vyas and M. Carlson, 2002 Snf1 protein kinaseand the repressors Nrg1 and Nrg2 regulate FLO11, haploid inva-sive growth, and diploid pseudohyphal differentiation. Mol. Cell.Biol. 22: 3994–4000.

Lengeler, K. B., R. C. Davidson, C. D’Souza, T. Harashima, W. C.Shen et al., 2000 Signal transduction cascades regulating fungaldevelopment and virulence. Microbiol. Mol. Biol. Rev. 64: 746–785.

Madhani, H. D., and G. R. Fink, 1997 Combinatorial control re-quired for the specificity of yeast MAPK signaling. Science 275:1314–1317.

Madhani, H. D., C. A. Styles and G. R. Fink, 1997 MAP kinaseswith distinct inhibitory functions impart signaling specificity dur-ing yeast differentiation. Cell 91: 673–684.

Martinez-Espinoza, A. D., J. Ruiz-Herrera, C. G. Leon-Ramirez

and S. E. Gold, 2004 MAP kinase and cAMP signaling pathwaysmodulate the pH-induced yeast-to-mycelium dimorphic transi-tion in the corn smut fungus Ustilago maydis. Curr. Microbiol.49: 274–281.

Mosch, H. U., R. L. Roberts and G. R. Fink, 1996 Ras2 signals viathe Cdc42/Ste20/mitogen-activated protein kinase module toinduce filamentous growth in Saccharomyces cerevisiae. Proc.Natl. Acad. Sci. USA 93: 5352–5356.

Mosch, H. U., E. Kubler, S. Krappmann, G. R. Fink and G. H. Braus,1999 Crosstalk between the Ras2p-controlled mitogen-activatedprotein kinase and cAMP pathways during invasive growth of Sac-charomyces cerevisiae. Mol. Biol. Cell 10: 1325–1335.

Obara, Y., A. M. Horgan and P. J. Stork, 2007 The requirement ofRas and Rap1 for the activation of ERKs by cAMP, PACAP, andKCl in cerebellar granule cells. J. Neurochem. 101: 470–482.

Olson, K. A., C. Nelson, G. Tai, W. Hung, C. Yong et al., 2000 Tworegulators of Ste12p inhibit pheromone-responsive transcriptionby separate mechanisms. Mol. Cell. Biol. 20: 4199–4209.

Palecek, S. P., A. S. Parikh and S. J. Kron, 2002 Sensing, signallingand integrating physical processes during Saccharomyces cerevi-siae invasive and filamentous growth. Microbiology 148: 893–907.

Pan, X., and J. Heitman, 1999 Cyclic AMP-dependent protein ki-nase regulates pseudohyphal differentiation in Saccharomycescerevisiae. Mol. Cell. Biol. 19: 4874–4887.

Pan, X., and J. Heitman, 2002 Protein kinase A operates a molec-ular switch that governs yeast pseudohyphal differentiation.Mol. Cell. Biol. 22: 3981–3993.

Pan, X., T. Harashima and J. Heitman, 2000 Signal transductioncascades regulating pseudohyphal differentiation of Saccharomy-ces cerevisiae. Curr. Opin. Microbiol. 3: 567–572.

Pearson, R. B., and B. E. Kemp, 1991 Protein kinase phosphoryla-tion site sequences and consensus specificity motifs: tabulations.Methods Enzymol. 200: 62–81.

Rispail, N., D. M. Soanes, C. Ant, R. Czajkowski, A. Grunler et al.,2009 Comparative genomics of MAP kinase and calcium-calcineurin signalling components in plant and human patho-genic fungi. Fungal Genet. Biol. 46: 287–298.

Roberts, R. L., and G. R. Fink, 1994 Elements of a single MAP ki-nase cascade in Saccharomyces cerevisiae mediate two develop-mental programs in the same cell type: mating and invasivegrowth. Genes Dev. 8: 2974–2985.

Robertson, L. S., and G. R. Fink, 1998 The three yeast A kinaseshave specific signaling functions in pseudohyphal growth. Proc.Natl. Acad. Sci. USA 95: 13783–13787.

Rupp, S., E. Summers, H. J. Lo, H. Madhani and G. Fink, 1999 MAPkinase and cAMP filamentation signaling pathways converge onthe unusually large promoter of the yeast FLO11 gene. EMBO J.18: 1257–1269.

Saxena, M., S. Williams, K. Tasken and T. Mustelin,1999 Crosstalk between cAMP-dependent kinase and MAP ki-nase through a protein tyrosine phosphatase. Nat. Cell Biol. 1:305–311.


Schneper, L., K. Duvel and J. R. Broach, 2004 Sense and sensibil-ity: nutritional response and signal integration in yeast. Curr.Opin. Microbiol. 7: 624–630.

Taxis, C., and M. Knop, 2006 System of centromeric, episomal, andintegrative vectors based on drug resistance markers for Saccha-romyces cerevisiae. Biotechniques 40: 73–78.

Tedford, K., S. Kim, D. Sa, K. Stevens and M. Tyers, 1997 Regulationof the mating pheromone and invasive growth responses in yeast bytwo MAP kinase substrates. Curr. Biol. 7: 228–238.

Toda, T., I. Uno, T. Ishikawa, S. Powers, T. Kataoka et al., 1985 Inyeast, RAS proteins are controlling elements of adenylate cyclase.Cell 40: 27–36.

Toda, T., S. Cameron, P. Sass, M. Zoller and M. Wigler,1987 Three different genes in S. cerevisiae encode the catalyticsubunits of the cAMP-dependent protein kinase. Cell 50: 277–287.

Truckses, D. M., J. E. Bloomekatz and J. Thorner, 2006 The RAdomain of Ste50 adaptor protein is required for delivery of Ste11to the plasma membrane in the filamentous growth signalingpathway of the yeast Saccharomyces cerevisiae. Mol. Cell. Biol.26: 912–928.

Truckses, D. M., L. S. Garrenton and J. Thorner, 2004 Jekyll andHyde in the microbial world. Science 306: 1509–1511.

Uesono, Y., M. P. Ashe and E. A. Toh, 2004 Simultaneous yet inde-pendent regulation of actin cytoskeletal organization and trans-lation initiation by glucose in Saccharomyces cerevisiae. Mol.Biol. Cell 15: 1544–1556.

Verstrepen, K. J., and F. M. Klis, 2006 Flocculation, adhesion andbiofilm formation in yeasts. Mol. Microbiol. 60: 5–15.

Vyas, V. K., S. Kuchin and M. Carlson, 2001 Interaction of the re-pressors Nrg1 and Nrg2 with the Snf1 protein kinase in Saccharo-myces cerevisiae. Genetics 158: 563–572.

Vyas, V. K., S. Kuchin, C. D. Berkey and M. Carlson, 2003 Snf1kinases with different beta-subunit isoforms play distinct rolesin regulating haploid invasive growth. Mol. Cell. Biol. 23:1341–1348.

Wach, A., A. Brachat, R. Pohlmann and P. Philippsen,1994 New heterologous modules for classical or PCR-basedgene disruptions in Saccharomyces cerevisiae. Yeast 10: 1793–1808.

Wach, A., A. Brachat, C. Alberti-Segui, C. Rebischung andP. Philippsen, 1997 Heterologous HIS3 marker and GFP re-porter modules for PCR-targeting in Saccharomyces cerevisiae.Yeast 13: 1065–1075.

Zappacosta, F., M. J. Huddleston, R. L. Karcher, V. I. Gelfand,S. A. Carr et al., 2002 Improved sensitivity for phosphopeptidemapping using capillary column HPLC and microionspray massspectrometry: comparative phosphorylation site mapping fromgel-derived proteins. Anal. Chem. 74: 3221–3231.

Zeitlinger, J., I. Simon, C. T. Harbison, N. M. Hannett, T. L.Volkert et al., 2003 Program-specific distribution of a tran-scription factor dependent on partner transcription factor andMAPK signaling. Cell 113: 395–404.

Communicating editor: O. Cohen-Fix


Systematic Epistasis Analysis of the Contributions of ...performed comprehensive epistasis analysis...

Documents

Transcript of Systematic Epistasis Analysis of the Contributions of ...performed comprehensive epistasis analysis...